基于四元数特定颜色对边缘检测的车牌定位

License plate location based on quaternion specific color-pair edge detection

车牌定位是车牌识别系统中的一个关键问题。提出了一种基于四元数特定颜色对边缘检测的车牌定位算法。首先将输入彩色图像用同色调全饱和度四元数形式表示;然后借助四对模板检测特定颜色对边缘,结合形态学膨胀运算提取潜在车牌区域;最后根据车牌形状约束条件定位车牌区域。该方法综合利用了车牌的颜色、边缘和形状特征,具有较好的鲁棒性。对各种情况拍摄的485幅车牌图像应用该算法,查全率达到96.8%,查准率超过93.2%。

License plate location plays an important role in the License Plate Recognition （LPR） system. A license plate location algorithm was presented, which was based on the quaternion specific color-pair edge detection technique. The original color image was first represented by using quaternion Same-Hue-Full-Saturation （SHFS） form. Then, four pairs of masks were used to detect specific color-pair edges. Next, mathematical morphology dilation operation was applied to extract potential license plate regions. Finally, several shape constraint conditions were employed to locate true license plate regions. The proposed method combined the color feature, edge feature and shape feature of license plates, and offered high robustness. The experimental results on 485 car images that are taken under various conditions show that the recall rate is 96.8% and the precision rate is more than 93.2%.